NEAR FUTURES: ASSOCIATIVE ARCHETYPES

摘要

Most computational design systems follow a Structuralist approach: a brief is fragmented into subcategories whose constrained parameters are resolved through predefined procedures. Adding hierarchies and iterations produces a template-like process that can be transferred to any similar task. This methodology is best represented by the field of knowledge-based modelling, which attempts to automate design processes by explicitly specifying all elements of a known and logical domain, its subcategories and every possible decision-rule for their combination. Like an expert system, all design problems are assumed to contain a linear path to an optimum solution. Techniques of self-organisation and bottom-up systems provide an alternative to solution-driven automation. Yet, they still require an explicit definition of parametric constraints and rules for generating outcomes that fit an intended architectural result. Most of these results are rooted in building sector domains and derived from functional specifications, ignoring the organisation of space from empirical knowledge. It is possible for space planning to exploit implicit human experiences of space as design drivers through a kind of computational intuition, but this requires new models that recognise spatial domains autonomously through data associations not aligned with key-performance and compliance indicators identified by building regulations.